OUP user menu

Still without impact on adverse post-operative outcomes: pre-operative statin therapy in patients undergoing cardiac surgery: reply

Oliver J. Liakopoulos, Gernot Wassmer, Thorsten Wahlers
DOI: http://dx.doi.org/10.1093/eurheartj/ehn354 2444-2445 First published online: 1 August 2008

We appreciate the authors' comments and thank them for emphasizing several limitations that must be considered for accurate interpretation of the reported statin treatment effects in our systematic review. As correctly stated in their comment, our analysis identified an unequal distribution of potentially confounding factors among patient groups, including a significantly higher prevalence of a β-blocker and aspirin therapy in statin pre-treated patients (P < 0.001). In large observational trials, the use of both β-blocker and aspirin has been associated with decreased mortality and morbidity following cardiac surgery.1,2 The importance of this treatment bias with unknown impact on the presented results is highlighted in detail in the limitation section and must be taken seriously into account when interpreting the results of our meta-analysis.

Furthermore, assessment of odds ratios (ORs) was performed after a detailed review and extraction of all available raw data from the included studies. The unadjusted ‘crude’ OR was determined when complete data sets from risk-adjusted treatment groups were not reported in the respective studies. Given the fact that adjustment for covariates and unmeasured confounders differ between included observational studies, our meta-analysis accounted for these discrepancies by analysing the prevalence of various perioperative variables and by clearly presenting the unequal distribution of potential confounding factors among statin-treated and untreated patients in a separate table. Most importantly, the implications of this treatment bias on the reported treatment effects of analysed outcomes were adequately discussed to ensure valid interpretation of our results. Nevertheless and following the authors' suggestion, a recalculation of the pooled OR derived by combining unadjusted and adjusted ORs from all included studies was performed (see Supplementary material online). This supplemental analysis demonstrates a 25% reduction for the endpoint, early all-cause mortality in statin pre-treated patients [OR 0.75; 95% confidence interval (CI): 0.63–0.89; P = 0.001 for overall effect], thereby indicating a reduced but still significant statin effect on early all-cause mortality when compared with the reported crude OR (OR 0.57; 95% CI: 0.49–0.67; P < 0.00001).

In light of the aforementioned limitations, we remain to our conclusion that our meta-analysis of predominantly observational trials strengthens the evidence for a preoperative statin use for prevention of adverse postoperative outcomes in patients undergoing cardiac surgery. However, it is important to re-emphasize that the results from our study should not be taken as an endorsement for an empiric perioperative statin therapy until conclusive data are accumulated from future randomized, controlled trials. Nevertheless, we strongly believe that it is appropriate to advocate an intensified perioperative statin therapy in all hyperlipaemic patients with coronary artery disease and multiple cardiac risks scheduled for heart surgery. In contrast to the author's final comment, our conclusions are based on data from several observational trials demonstrating a decrease in early postoperative mortality or morbidity in statin-treated patients that persisted even after multivariate analysis and risk adjustment of patient groups.3,4 Additional supporting evidence is provided by two recent reports showing that statin withdrawal in the immediate postoperative period is independently associated with an increased risk for late in-hospital mortality5 and that statins reduce the risk of major adverse cardiac events in hyperlipaemic patients undergoing coronary artery bypass grass surgery.6